Regulation and Foodborne Disease Outbreaks

As noted below, in response to this column by Paul Krugman on the failure of food safety regulation in recent years, questions have been raised about whether food borne disease outbreaks have been increasing or decreasing. Here's Alex Tabarrok at Marginal Revolution:

Krugman gets a Rotten Tomato, by Alex Tabarrok: Paul Krugman is attacking Milton Friedman (again) for rotten tomatoes...:

Lately, however, there always seems to be at least one food-safety crisis in the headlines — tainted spinach, poisonous peanut butter and, currently, the attack of the killer tomatoes.

How did America find itself back in The Jungle?

I was curious so I collected data from the Center for Disease Control on Foodborne Disease Outbreaks from 1998-2006. The data only go back to 1998 because in that year the CDC changed its surveillance system creating a discontinuity but note that we are covering a chunk of the Clinton years and are well within the time frame over which Krugman says the safety system has degenerated. Here's the result:

What we see is a lot of variability from year to year but a net downward trend. You can also look at cases per year which are more variable but also show a net downward trend. No evidence whatsoever that we are back "in The Jungle."

Mathew Yglesias says:

...isn't the pattern of spiking in even years and declining in odd years sort of strange...

I am going to turn this over to a food economist the the U.S. Food Policy Blog:

Marginal Revolution covers food safety, by Parke Wilde: Alex Tabarrok at Marginal Revolution gives Paul Krugman a hard rap on the knuckles for criticizing federal food safety regulators without noting the relevant data. Krugman's recent column describes a crisis of foodborne illness outbreaks, while Tabarrok points out CDC statistics showing that at least some kinds of outbreaks haven't increased. Still, the CDC data aren't the whole story. Here is my comment:

...I try to ignore the ideologues. On the one hand, Grover Norquist and Milton Friedman have hopes for market incentives that are not only unrealistic, they are very far from the current state of good contemporary economic thought on food safety.

On the other hand, I give little weight to critics who hold out unrealistic expectations for government prevention of all food safety risks, not so much because they care about food safety, but because they just have large ambitions for government interventions in the economy.

Current economic thought draws on a fairly specific diagnosis of different kinds of imperfect information, each of which calls for a different policy response, which can range from laissez faire to labeling to testing to regulation.

A standard mainstream -- or even somewhat conservative -- text on the economics of food safety is John Antle's book on food safety. It acknowledges market failures in food safety, and particularly recognizes the need for strong government regulation of food-borne pathogens, because in many cases consumers cannot recognize the safety of the food even after purchase, and hence cannot defend their own interests in the marketplace. In my reading, ... on foodborne disease [Antle] seems to me closer to Krugman than Friedman.

There are some exciting private market innovations in food safety recently. For example, the buyers for major supermarket chains are getting more sophisticated in demanding safety from their suppliers, which allows the market to achieve good outcomes that individual consumers could not command on their own.

At the same time, it is fair to say USDA and FDA oversight of food safety have fallen far short of a balanced position. The CDC stats on outbreaks have a number of shortcomings, and don't suffice to make me think otherwise. Take something like the USDA's refusal to let Creekstone beef voluntarily test its own product for BSE. ... The so-called "regulators" are way out of step with the public interest position of economists, even market economists who appreciate the market's accomplishments on its good days.

Let me take this a step further. Here are the actual data from the individual reports from the CDC (the totals are shown in the graph from Alex's post):

	1998	1999	2000	2001	2002	2003	2004	2005	2006
Bacterial	257	222	226	235	226	196	208	188	223
Chemical	48	32	37	52	46	54	47	40	53
Multiple		2	3	7	12	7	5	6
Parasitic	4	3	6	5	5	3	8	6	9
Viral	59	109	176	156	205	149	251	170	337
Confirmed	368	368	448	455	494	409	519	410	623
Suspected									275
Unknown	946	976	969	783	838	663	800	572	349
Total	1314	1344	1417	1238	1332	1072	1319	982	1247

When I look at these data, it's hard to find any consistent trends. The increase in confirmed cases appears, for the most part, to be due to an increase in confirmed viral outbreaks, but it's hard to know whether the increase is more actual cases, which would be worrisome, or just better technology at identifying them. For the rest of the confirmed cases, there is not that much of a pattern. But no matter what the trends in confirmed cases say, with more than half the cases due to unknown causes and so much variation in the number of unknown etiologies, it's hard to know how to interpret the overall figures shown in the graph above as a rebuttal to Krugman.

Stepping away from the data for a moment, and given their quality that may be wise, the larger issue is whether the government should intervene and regulate food markets to promote safety, or follow Friedman's advice and let the private sector take care of the problem. The market failure due to asymmetric information identified above, and the advice of experts in the area make a compelling case for regulation.

Update: Tyler Cowen, Alex Tabarrok, and others have asked in comments that I update the post to say that these data do not support Krugman. Yes, that's true, but it's because these data are too noisy to support anything (they suffer from a huge errors in variable problem that appears to be time-varying, and also appear to be heteroskedastic). These data are of such poor quality that you can't conclude anything. The trend could be rising, it could be flling, it could be constant, but these data can't tell us the answer. So, the rebuttal  argument seems to be as follows. Someone (Krugman) makes a claim. To check the claim, I go get really, really noisy data, run a regression, and say, "see - the hypothesis doesn't stand up to the data!" That doesn't seem very compelling to me.